Kohesi: Jurnal Sains dan Teknologi

Articles

Vol. 10 No. 8 (2026): Kohesi: Jurnal Sains dan Teknologi, ISSN 3025-1311

PERBANDINGAN PERFORMA ARSITEKTUR MODEL U-NET DAN DEEPLAB UNTUK SEGMENTASI LOW GRADE GLIOMA (LGG) BERBASIS CITRA MAGNETIC RESONANCE IMAGING (MRI)

Submitted
January 6, 2026
Published
2026-01-06

Abstract

Brain tumor segmentation in Magnetic Resonance Imaging (MRI) is an essential step in supporting accurate diagnosis and appropriate medical treatment planning. This study aims to compare the performance of U-Net and DeepLabV3 models in segmenting Low Grade Glioma (LGG) brain tumors based on MRI images. The dataset used in this research was obtained from Kaggle, consisting of 20 selected datasets, including 10 datasets with large tumors and 10 datasets with small tumors. The data were divided into training and testing sets with an 80:20 ratio. The training process was conducted by applying data augmentation techniques, such as resizing and normalization, and optimizing the models using the Adamax algorithm. On the LGG 1 (Big Tumor dataset), DeepLabV3 achieved a Dice Coefficient of 89.60% and an IoU of 81.25% on the testing data, while U-Net achieved a Dice Coefficient of 86.60% and an IoU of 76.49%. On the LGG 2 (Small Tumor dataset), DeepLabV3 also demonstrated superior performance with a Dice Coefficient of 88.14% and an IoU of 79.47%, whereas U-Net only reached a Dice Coefficient of 55.95% and an IoU of 39.84%. These findings indicate that DeepLabV3 is more effective in multi-scale segmentation, especially for small tumor objects, compared to U-Net. Further development is recommended to utilize larger and more diverse datasets, and to integrate the model into medical decision support systems to enhance the effectiveness of brain tumor diagnosis

References

  1. [1] E. Lyutfi, R. Georgieva, G. Stoyanov, and D. Dzhenkov, “Tumor growth patterns in central nervous system tumors with astrocytic differentiation,” Glioma, vol. 3, no. 2, p. 67, 2020, doi: 10.4103/glioma.glioma_8_20.
  2. [2] S. Mueller and M. A. Karajannis, “Pediatric Brain Tumors – High-Grade Glioma,” in Swaiman’s Pediatric Neurology (Sixth Edition), 2017, pp. 979–984. doi: https://doi.org/10.1016/B978-0-323-37101-8.00126-0.
  3. [3] H. A. Hafeez et al., “A CNN-Model to Classify Low-Grade and High-Grade Glioma From MRI Images,” IEEE Access, vol. 11, no. April, pp. 46283–46296, 2023, doi: 10.1109/ACCESS.2023.3273487.
  4. [4] W. Aiman, Low grade gliomas. 2023. doi: https://www.ncbi.nlm.nih.gov/books/NBK560668/.
  5. [5] The Cancer Genome Atlas Research Network, “Comprehensive, Integrative Genomic Analysis of Diffuse Lower-Grade Gliomas,” N. Engl. J. Med., vol. 372, no. 26, pp. 2481–2498, 2015, doi: 10.1056/nejmoa1402121.
  6. [6] H. Du, C. Piyawajanusorn, G. Ghislat, and P. J. Ballester, “Predicting temozolomide response in low-grade glioma patients with large-scale machine learning,” bioRxiv, 2024, doi: https://doi.org/10.1101/2024.12.22.629995.
  7. [7] B. H. Menze et al., “The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS),” IEEE Trans. Med. Imaging, vol. 34, no. 10, pp. 1993–2024, 2015, doi: 10.1109/TMI.2014.2377694.
  8. [8] O. Ronneberger, P. Fische, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” in Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015, 2015, pp. 234–241. doi: https://doi.org/10.1007/978-3-319-24574-4_28.
  9. [9] L. C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, “DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 40, no. 4, pp. 834–848, 2018, doi: https://doi.org/10.1109/TPAMI.2017.2699184.
  10. [10] D. Xu, X. Zhou, X. Niu, and J. Wang, “Automatic segmentation of low-grade glioma in MRI image based on UNet++ model,” J. Phys. Conf. Ser., vol. 1693, no. 1, 2020, doi: 10.1088/1742-6596/1693/1/012135.
  11. [11] Q. Zhang, W. Qi, H. Zheng, and X. Shen, “CU-Net: A U-Net Architecture for Efficient Brain-Tumor Segmentation on BraTS 2019 Dataset,” 2024 4th Int. Conf. Mach. Learn. Intell. Syst. Eng. MLISE 2024, pp. 255–258, 2024, doi: 10.1109/MLISE62164.2024.10674119.
  12. [12] I. Aboussaleh, J. Riffi, K. El Fazazy, M. A. Mahraz, and H. Tairi, “Efficient U-Net Architecture with Multiple Encoders and Attention Mechanism Decoders for Brain Tumor Segmentation,” Diagnostics, vol. 13, no. 5, 2023, doi: 10.3390/diagnostics13050872.
  13. [13] S. Saluja, M. C. Trivedi, and S. S. Sarangdevot, “Advancing glioma diagnosis: Integrating custom U-Net and VGG-16 for improved grading in MR imaging,” Math. Biosci. Eng., vol. 21, no. 3, pp. 4328–4350, 2024, doi: 10.3934/mbe.2024191.

Similar Articles

1-10 of 56

You may also start an advanced similarity search for this article.